| Figure 12.1 - AdaBoost cross-validation performance.png | 188.3 KB | |
| Figure 12.10 - Average and cumulative returns by factor quantile.png | 201.3 KB | |
| Figure 12.11 - Rolling IC for 1-day and 21-day return forecasts.png | 183.2 KB | |
| Figure 12.12 - LightGBM feature importance.png | 102.7 KB | |
| Figure 12.13 - Partial dependence plots for scikit-learn GradientBoostingClassifier.png | 369.1 KB | |
| Figure 12.14 - Partial dependence as a 3D plot.png | 1.1 MB | |
| Figure 12.15 - SHAP summary plots.png | 92.8 KB | |
| Figure 12.16 - SHAP force plot.png | 22.8 KB | |
| Figure 12.17 - SHAP clustered force plot.png | 67.9 KB | |
| Figure 12.18 - SHAP interaction plot.png | 24.2 KB | |
| Figure 12.19 - Strategy performance—cumulative returns and rolling Sharpe ratio.png | 249.9 KB | |
| Figure 12.2 - The gradient boosting algorithm.png | 46 KB | |
| Figure 12.20 - Information coefficient for high-frequency features.png | 92 KB | |
| Figure 12.21 - Average 1-min returns and cumulative returns by decile.png | 219 KB | |
| Figure 12.3 - Cross-validation performance of the scikit-learn gradient boosting classifier.png | 179.2 KB | |
| Figure 12.4 - Hyperparameter impact for the scikit-learn gradient boosting model.png | 250.2 KB | |
| Figure 12.5 - Impact of the gradient boosting model hyperparameter settings on test performance.png | 261.9 KB | |
| Figure 12.6 - Depth-wise vs leaf-wise growth.png | 20.9 KB | |
| Figure 12.7 - Predictive performance and runtimes of the various gradient boosting models.png | 74.7 KB | |
| Figure 12.8 - Overall and daily IC for the LightGBM and CatBoost models over three prediction horizons.png | 98.7 KB | |
| Figure 12.9 - Coefficient estimates and their confidence intervals for different forecast horizons.png | 183.6 KB | |
